Andy,
I used that program when I did my 500 kHz exciter a few years ago. At
the time, I went back to some old source material at work, and compared
the hand-calculated results with the program. The two produced the same
results, so the program is certainly OK.
You do raise an excellent point with component tolerances, however. For
that filter, I used a very accurate Phillips digital RLC meter at work
to select capacitors from a batch of 2% units that I had in stock. Armed
with the measured capacitance, I hand-selected resistors from ample
stocks of "close" 1% values to equal the reactance of the capacitors at
the break points. The 10k0 resistors were already 0.1%, so I didn't
bother trying to grade them out.
The resulting plot of phase vs. frequency may be seen at:
http://www.w1tag.com/pix/600M_Phase.jpg
As you noted, it does have a problem at the high end, probably due to
the op-amp input capacitance, which was not considered. A mental note
was made at the time!
The only schematic I can find is hand drawn, but using the program, the
parameters were F1=100Hz, F2=3500Hz, C=10.0nF and 5 filters (per side,
for a total of 10 op-amp sections. I tweaked the cap values to only use
100, 10 and 1 nF. The critical thing in tweaking is to maintain those
breakpoint frequencies that represent Xc=R.
John, W1TAG
On 11/7/2010 3:46 PM, Andy Talbot wrote:
That's a neat bit of software!
But I've noticed in all the all-pass design procedures, there
are choices of optimising, Monte Carloe-ing, adjusting and tuning the
RC part of the networks. But none seem to take into account the effect
of mismatch of the nominally matched pair of resistors around each
opamp. Or of residual amplitude mismatching.
But it does seem that if you are prepared to spend the time getting all
the Rs and Cs within a micro percent of the correct values, then apply
the same effort to amplitude and phase matching the channels, a quite
phenomenal performance wrt. sideband suppression is theoretically possible.
But I guess in any practical scenario, most constructors will tweak for
an acceptable performance and leave it alone. For upconverting at LF,
we're only interested in narrow band signals - the entire WSPR segment
is only 200kHz wide - and a simple 2 stage (one opamp in each of I and
Q) is adequate. In that case a pair of preset or variable resistors can
be tweaked quite quickly. 40 - 50dB supression seems almost guaranteed
over a 100Hz using a predictable mixers like fast CMOS switches, with a
comparable level of carrier suppression. I've built two now, one with
FST3125 and a balanced output transformer, and one with a dual quad
switch, FST3257. The latter gives somewhat better performance, but is
slightly more complex.
Andy
www.g4jnt.com <http://www.g4jnt.com>
On 7 November 2010 19:33, Johan H. Bodin <[email protected]
<mailto:[email protected]>> wrote:
Andy,
here's another nice program for all-pass I/Q filter design:
http://www.tonnesoftware.com/quad.html
73
Johan SM6LKM
----
Andy Talbot wrote 2010-11-06 14:59:
> Very Interesting.
> Where did you find the design data for the allpass network?
I've only been
> able to find the values for up to 3+3 allpass stages.
>
> Mind you, at that number of chips a DSPic begins to look attractive.
>
>
> Andy
> www.g4jnt.com <http://www.g4jnt.com/> <http://www.g4jnt.com
<http://www.g4jnt.com/>>
>
>
> On 6 November 2010 11:52, <[email protected]
<mailto:[email protected]> <mailto:[email protected]
<mailto:[email protected]>>>
> wrote:
>
> Nothing earth shattering ... another way to skin the cat. The
137 and 500
> kHz units have been in use for a few years ... just now
getting around to
> the write up.
>
> http://www.w1vd.com/137-500Phasingexciter.html
>
> Jay W1VD WD2XNS WE2XGR/2
>
>
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.5.449 / Virus Database: 271.1.1/3242 - Release Date: 11/07/10
07:34:00
|
